jack-field.com

Jack Field

Software engineer and angel investor, now doing my MBA at Berkeley Haas. I'm moving from engineering into product, building on years of turning customer discovery into products people pay for. Before Haas I ran engineering at Gecko Risk and spent six years at Quant Insight, where I built a risk model now on Goldman Sachs Marquee.

githublinkedinemailDownload resume (PDF)

The kitchen · right now

  • foreman

    Head chef

    asleep

  • the Gate

    On the pass

    asleep

  • Jack

    Owner

    asleep

  • Product

    Culinary Director

    asleep

0 on shift

live

This is live. Each character is part of the system that builds my projects: foreman hands out tasks to AI agents, a second agent reviews their code, and I appear whenever I'm working too. A sleeping character has nothing to do right now.

The log · what went out

  1. 29 Octrolodeck-aiShow a Kept company's News and Events on its Watchlist card
  2. 29 Octrolodeck-aiShow past Events newest first, at most 30, above the upcoming ones
  3. 29 Octrolodeck-aiKeep every history-search article, and show a company's own post when it is news
  4. 29 OctcaddieState when Strengths and results were last read, and flag a stale derive
  5. 29 Octrolodeck-aiShow only startup Events in the Diary: drop earnings calls and media events unless a Kept company attends
  6. 29 OctcaddieCount the coverage record by edition, and name each event's season
  7. 29 Octrolodeck-aiDeal, Watchlist and group News once per company, not once per Source
  8. 29 OctforemanAnnounce the Scout to jack-field while it runs
  9. 26 OctcaddiePin the Player column and drop the inner scroller on narrow screens
  10. 26 OctcaddieState the answer in a sentence above the Course Profile
  11. 26 OctcaddieSay why a Player is unscored: the Course, the Weightings, or their record
  12. 26 Octrolodeck-aiRank the Deck the same whether a User picks every Stage or none
  13. 25 Octrolodeck-aiAdd TechCrunch feeds to News
  14. 25 Octrolodeck-aiCapture TechCrunch's main and Y Combinator feeds by hand, for #221
  15. 25 OctcaddieLeave Players unranked when the Course asks nothing that separates them
  16. 25 OctcaddieGive hole pars their own check, so the par mix survives untrusted yardages
  17. 25 OctcaddieShow the Record here column only when some Player has a venue record value
  18. 25 OctcaddieRead the schedule's Date as the final day, and keep a Tournament next until it ends
  19. 25 Octrolodeck-aiMake the Deck card swipeable by pointer drag
  20. 25 Octrolodeck-aiLet a User Pass a Kept Profile from the Watchlist
  21. 25 Octrolodeck-aiClose a freshness alarm once its Source is running again
  22. 25 Octrolodeck-aiShow Sectors and Stages as words, not as slugs
  23. 25 OctforemanKeep a Pitch open at least twelve hours before it can expire
  24. 24 Octrolodeck-aiReplace Techmeme in News with Launch HN

Merged pull requests, newest first. Most were written by one agent and reviewed by another, and those merge only once review and CI both pass.

What I am building

foreman

Turns a GitHub issue into a merged pull request, unattended.

I write a ticket with clear acceptance criteria. foreman picks it up, starts a coding agent in its own isolated copy of the repo, and runs up to three at a time. Each agent opens a pull request and a second agent reviews it. If review or CI rejects the work, foreman sends it back with the findings, and if it runs out of things it is allowed to decide, it stops and asks me. Once a day it also reads each repo and pitches three things worth fixing or building next.

fieldjw96/foreman

rolodeck-ai

A swipeable deck of early-stage Bay Area startups.

The best early-stage companies are hard to spot, because the signals are scattered across accelerator batches, SEC filings and launch posts on Hacker News. Rolodeck pulls them into one deck, ranked by what you care about, and deals them one at a time. Swipe to keep or pass. Companies you keep go on a watchlist with their news and upcoming events alongside, whether they're a possible internship, worth an introduction, or just one to watch.

fieldjw96/rolodeck-ai

What I have built

  1. The Gecko Risk platform

    Gecko Risk, 2025 to 2026

    The web application at the centre of an electric vehicle risk data company: the thing its customers actually work in. Built in-house to replace Tableau, and it helped win Tesla and three insurers.

    Grew ARR 35% and took churn from 20% to zero.

  2. The Qi equity risk model

    Quant Insight

    A macro risk model for equities, and one of only four risk models offered on the Goldman Sachs Marquee data platform. It drives roughly half of Quant Insight's annual recurring revenue.

    On Goldman Sachs Marquee.

  3. iQ

    Quant Insight

    A personalised trading assistant that dropped into a client's existing workflow rather than asking them to leave it. At its height it was used by more than a hundred institutional clients, and was then repurposed as a retail product and spun out.

    Spun out of Qi.

Before this

Aug 2025

Aug 2026

London

Gecko RiskElectric vehicle risk data

Head of Engineering

  • Owned product and engineering for the customer-facing analytics platform. Convinced the CEO and CPO to replace Tableau with an in-house build. It won Tesla and three insurers as clients and grew ARR 35%.
  • Replaced an outsourced dev team with AI agents that triage alerts, draft fixes and ship features, after a two month benchmark. Platform costs fell 96% and dashboards shipped in three days instead of three weeks.
  • Took churn from 20% to zero by building an in-product feedback tool and using it to scope an AI charting feature, so customers could generate any chart from a plain-language question rather than request a bespoke dashboard.
  • Consolidated a fragmented .NET codebase into a modular, serverless Python ETL platform on AWS that any engineer, or agent, could pick up and extend. It runs eight times faster.

Oct 2024

Aug 2025

London

RVUConfused.com, Uswitch, Money, Tempcover

Software Engineer

  • Ran discovery with engineering and marketing across all four brands and turned it into a single user-tracking spec and roadmap, with rollout ordered by traffic.
  • Shipped the customer data product at roughly a million users a month with tenfold tracking coverage. It powered cross-brand campaigns that lifted click-through rate 6%.

Jul 2022

Jul 2023

London

TechFieldTech consultancy

Founder

  • Turned inbound requests from founders in my network into a paid tech consultancy.
  • Designed and built technical products and ETL pipelines for startups, most notably a generative AI marketing content tool for Pockla, plus advising the founders on cloud architecture.
  • Data analytics lead with OPUS, turning insights on exceptional young leaders into signals their VC arm could source against.

Sep 2018

Sep 2024

London

Quant InsightFinancial data analytics

Intern → Quant Analyst → Senior Data Scientist → Tech Lead

  • Ran discovery with hedge funds and data providers to find the gap in macro risk modelling, then derived the model and wrote the whitepaper. It is one of four risk models on the Goldman Sachs Marquee platform and drives half of Qi's annual recurring revenue.
  • Eighth hire and second technical one, promoted three times. Wrote machine learning trading signals that back-tested above 20% annualised out of sample, which became the product behind deals with Squared Financial and Interactive Investor.
  • Conceived and launched a personalised trading assistant that dropped into clients' existing workflow. Over a hundred institutional users at its height, then led its relaunch as a retail trader product.
  • As tech lead, set technical direction and managed a team of eight, balancing the roadmap against two data migrations and an infrastructure rebuild that cut compute cost 45%.

2014

2017

London

Venrex Investment ManagementVenture capital

Associate

  • Two stints. Evaluated early-stage startups as investments, visiting companies and assessing product, market and traction, alongside investor communications and account administration.
  • Automated investor and HMRC communications, removing twenty hours of manual work a month.

Education

  • Berkeley HaasAug 2026 to May 2028

    Berkeley, California

    MBA. Member of the Haas Tech, Product Management and AI clubs. Fourth in the Pegasus Airline x Berkeley Haas Hackathon.

  • University of BristolSep 2015 to Jun 2018

    Bristol

    Engineering Mathematics, BEng Hons, First Class. Top of the class, the equivalent of a 4.0.

Additional

Investing.
Angel investor in 14 early-stage startups since 2022, across tech and consumer, including Runna.
Stack.
Python, TypeScript, C#, SQL, Terraform, AWS (S3, Glue, Athena, Lambda, EC2, RDS, SQS), GCP (BigQuery, Dataform, Cloud Run), Claude Code, Cursor.
Achievements.
Gold in the Maths Olympiad. Represented East England in national hockey tournaments.
Interests.
Arsenal, golf, padel, and live music: more than fifty festivals and concerts.

Download resume (PDF)